Comprehensive Manual for SUV Computation and Digital Reference Objects for its Verification

Working version 2.0

Michael Vácha ^1,2,3, The image biomarker standardisation initiative (IBSI) $, Frank Hofheinz ³, Anja Braune ^3,4,5, Steffen Löck ^1,2,6,7,8, Alex Zwanenburg ^1,6

¹OncoRay-National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Helmholtz-Zentrum Dresden-Rossendorf, Dresden, Germany

²Helmholtz-Zentrum Dresden-Rossendorf, Institute of Radiooncology - OncoRay, Dresden, Germany

³Institute of Radiopharmaceutical Cancer Research, Helmholtz-Zentrum Dresden-Rossendorf, Dresden, Germany

⁴Department of Nuclear Medicine, University Hospital Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany

⁵Carl Gustav Carus Faculty of Medicine, Technische Universität Dresden, Dresden, Germany

⁶National Center for Tumor Diseases (NCT), NCT/UCC Dresden, a partnership between DKFZ, Faculty of Medicine and University Hospital Carl Gustav Carus, TUD Dresden University of Technology, and Helmholtz-Zentrum Dresden-Rossendorf (HZDR), Germany

⁷Department of Radiotherapy and Radiation Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany

⁸German Cancer Consortium, partner site Dresden, and German Cancer Research Center, Heidelberg, Germany

$ Participants and their affiliations are to be specified.

Introduction

There is an increasing interest in using positron emission tomography (PET) data for segmentation, biomarker identification, and outcome prediction. To make PET values comparable between patients, these values are often normalized to a standard unit, which is, in most cases, the body-weight-normalized standardized uptake value (SUV_bw). By definition, the SUV_bw can be computed from the formula:

\[ \text{SUV}_{\mathrm{bw}} = \frac{\scriptstyle A_c \, W}{\scriptstyle D} \]

where \({A}_{c}\) represents the measured activity concentration within a region of interest or voxel in Bq/ml, \(W\) the weight of the patient in g, and \(D\) the total administered radionuclide dose at the time to which the voxel values correspond, in Bq. However, PET imaging data in Digital Imaging and Communications in Medicine (DICOM) files are not stored as standardized uptake values. Consequently, the voxel values and associated metadata have to be correctly interpreted by the viewing or analytical tools in order to convert the voxel values to SUV_bw. In the radiomics field, many tools were not primarily designed for PET imaging and may not process all relevant metadata correctly. Consequently, identical images may be processed inconsistently across various software tools. This can give rise to two potential issues:

• incorrect or inconsistent SUV calculation, resulting in under- or overestimation of SUV_bw values, which may introduce unrecognized bias and reduce generalizability;
• errors that lead to patient exclusion and a decreased sample size.

By creating reference standards for SUV_bw computation by software tools, we aim to:

• verify that SUV_bw is computed correctly;
• maximize the number of PET imaging series that can be used for quantitative image analysis.

Therefore, we:

1. studied how PET images are stored in real-world imaging DICOM files;
2. summarized the instructions on how to read these files and calculate SUV_bw;
3. generated a set of representative digital reference objects (DROs) to assess whether imaging software can accurately interpret diverse PET DICOM formats.

Methods

Metadata analysis

To analyze how PET imaging data are stored, we searched for real-world PET human imaging series via two sources:

• The Cancer Imaging Archive Radiology Portal
• our internal imaging archives

We identified 37 cohorts containing relevant PET imaging data:

ACRIN-FLT-Breast, ACRIN-NSCLC-FDG-PET, Anti-PD-1_Lung, BREAST-DIAGNOSIS, CC-Tumor-Heterogeneity, CMB-CRC, CMB-GEC, CMB-LCA, CMB-MEL, CMB-MML, CMB-PCA, CPTAC-CM, CPTAC-HNSCC, CPTAC-LSCC, CPTAC-LUAD, CPTAC-PDA, CPTAC-SAR, CPTAC-UCEC, CT-vs-PET-Ventilation-Imaging, Head-Neck-PET-CT, Internal-Berlin, Internal-Dresden, Internal-Munich, Lung-PET-CT-Dx, NaF PROSTATE, NSCLC Radiogenomics, QIN-BREAST, RIDER Lung PET-CT, Soft-tissue-Sarcoma, TCGA-BLCA, TCGA-KIRP, TCGA-LUAD, TCGA-LUSC, TCGA-PRAD, TCGA-THCA, TCGA-UCEC, and VAREPOP-APOLLO.

We excluded imaging series without attenuation correction and detector normalization applied, as well as series with only one image within, maximum intensity projections and other reprojections. For all remaining series, we scanned a preselected set of attributes (listed in Suppl. Table 1) and analyzed how they were used in practice.

Manual for PET DICOM SUV_bw conversion

Based on literature (DICOM standards, QIBA consensus, Turku PET center manual, USZ technical note, our expertise, and the metadata analysis, we summarized the rules for converting PET images into SUV_bw-normalized images. Suppl. Fig. 1 includes a diagram summarizing the recommended strategy.

Construction of digital reference objects

Furthermore, we assembled a comprehensive set of digital reference objects (DROs) for verifying SUV conversion. All DRO DICOM files were synthesized de novo using Python pydicom library version 3.0.1. The design of these DROs was chosen to resemble common PET calibration phantoms – each DRO includes one hot sphere (SUV_bw = 4.00), one cold sphere (SUV_bw = 0.20), a background region (SUV_bw = 1.00), and surrounding zero-activity region (SUV_bw = 0.00) (See Fig. 1).

All DROs are identical in terms of the design, volumes, and resulting SUV_bw values, while the stored voxel values and DICOM attributes vary based on the intended use of each DRO. The region of interest (ROI) for SUV evaluation covers the whole DRO volume, excluding the surrounding region. For the sake of compatibility, it is provided in two formats: as a binary NIfTI (Neuroimaging Informatics Technology Initiative) mask, and as a radiotherapy structure set (RTSTRUCT) containing one 3D contour.

To verify a software tool computes the SUV_bw correctly, either:

1. directly compute SUV_{bw_max}, SUV_{bw_min}, and SUV_{bw_med} from the maximum, minimum, and median voxel value within the ROI;

2. convert the DRO DICOM file to SUV_bw and extract the DRO maximum, minimum, and median SUV_bw values from the ROI;

3. convert the DRO DICOM file to SUV_bw and visually inspect the values in the hot sphere, cold sphere, and background region.

All values should be calculated with a precision of two decimal digits. The target values are identical across all cases and are defined as follows:

• Hot sphere / maximum DRO value = 4.00 SUV_bw
• Cold sphere / minimum DRO value = 0.20 SUV_bw
• Background region / median DRO value = 1.00 SUV_bw

Fig. 1: Visualization of the DRO design in axial, coronal, and sagittal planes (from left to right). Voxel values are shown in SUVbw.

Results

Baseline characteristics

For the metadata analysis, we identified and examined 3106 PET imaging series from 1710 patients. These images were acquired by a large variety of PET scanner models, with the majority (92.8%) from three PET scanner manufacturers: 763 (24.6%) series were from Siemens (Siemens Healthineers, Erlangen, Germany), 1846 (59.4%) from GE (GE HealthCare, Chicago, USA) and 273 (8.8%) from Philips (Philips Healthcare, Amsterdam, Netherlands). The remaining scans were from CTI/CPS (CTI Molecular Imaging/CPS Innovations, Knoxville, USA) or the Manufacturer (0008,0070) attribute did not contain any name of a PET scanner manufacturer (Table 1).

Table 1: Percentage of series and a list of unique Manufacturer Model Name (0008,1090) values by scanner manufacturer.

Manufacturer	Perc	Models
CTI/CPS	6%	1023, 1024, 1062, 1080, 962
GE	59%	Advance, Discovery 610, Discovery 690, Discovery 710, Discovery IQ, Discovery LS, Discovery MI, Discovery MI DR, Discovery RX, Discovery ST, Discovery STE
Philips	9%	Allegro Body(C), GEMINI TF Big Bore, GEMINI TF TOF 16, Guardian Body(C), TruFlight Select
Siemens	25%	1080, 1093, 1094, Biograph 20_mCT, Biograph 64_mCT, Biograph Horizon, Biograph_mMR, Biograph128_mCT 4R, Biograph128_Vision 450 Edge, Biograph16_Horizon 3R, Biograph20_mCT, Biograph20_mCT 3R, Biograph40_mCT, Biograph40_mCT 4R, Biograph40_TruePoint, Biograph6_TruePoint, Biograph64_mCT, Biograph64_mCT 3R, Biograph64_Vision 600, SOMATOM Definition AS_mCT
Unknown	1%	DicomCleaner, Integrity Medical Image Importer

Rescale slope (m) and intercept (b)

Background

Due to the wide range of the measured voxel values in PET imaging data, the use of rescale slope (Rescale Slope; 0028,1053) ensures that stored values are within the range that can be stored in the voxel data type (e.g., signed 16-bit integers can store integer values from -32,768 to 32,767). Applying the rescale slope to the stored voxel data leads to conversion of the stored voxel values to the physical units, that are specified in Units (0054,1001). Independently of the units, it is a mandatory first step in SUV calculation:

\[ \text{U} = {m \, P + b} \]

where P is the stored voxel value (Pixel Data; 7fe0,0010), m the rescale slope, b the rescale intercept (Rescale Intercept; 0028,1052). The rescale slope may vary for each slice and must be applied on a slice-wise basis. Since the rescale intercept is required to be 0 in all PET studies, it can be omitted from the formula.

Metadata analysis

Comparing series, a wide range of rescale slopes was observed in the studied data. In general, the rescale slope values within each series were either:

• constant (N=1015, 32.7%)
• different in each slice (N=1698, 54.7%)
• rather constant with a few exceptions (N=391, 12.6%) - we noted the exceptions typically included slices with the highest uptake (e.g., slices containing the tumor region).

In our dataset, all PET imaging series had Rescale Intercept attribute present and its value was zero.

Digital reference object

The following DRO was constructed:

• DRO10 with multiple values of the Rescale Slope attribute.

  Possible issues:

  • Rescale Slope is not applied slice-wise

Recommendations

The Rescale Intercept (0028,1052) attribute must be checked to ensure it equals 0. Otherwise, SUV computation is not possible.

Justification: The Rescale Intercept is required to be present and equal to zero in PET images.

The Rescale Slope (0028,1053) attribute must be present and have a positive, non-zero value; otherwise, SUV computation is not possible. The corresponding Rescale Slope (0028,1053) has to be applied to all stored voxel values within a slice, independently of other parameters such as the units.

Justification: The Rescale Slope is required in PET images to convert voxel values to physical units and may vary between slices.

Voxel values (P) and their unit

Background

The attribute Pixel Data (7fe0,0010) stores the voxel values. As for other imaging modalities, the Pixel Data attribute can be read using other attributes from the DICOM Image Pixel Module, such as Rows (0028,0010), Columns (0028,0011), Bits Allocated (0028,0100), or Pixel Representation (0028,0103). While single-slice DICOM images still prevail, multi-planar DICOM images containing 3-D voxel data (i.e., ‘one-file DICOM’ formats) may become increasingly common in the future. For PET imaging, the interpretation of these values depends on the unit specified in the Units (0054,1001) attribute. As mentioned before, the attribute specifies to which physical units are the voxel values converted by applying the rescale slope. In general, the strategy can be divided into two pathways: A) direct conversion to SUV_bw; B) conversion to Bq/ml followed by conversion from Bq/ml to SUV_bw.

Units = “BQML”

The attribute Units (0054,1001) is commonly set to “BQML”, which signalizes physical units in becquerels per milliliter. Assuming W and D are accordingly corrected and expressed in proper units, U may be used directly in the formula:

\[ \text{SUV}_{\mathrm{bw}} = \frac{\scriptstyle U \, W}{\scriptstyle D} \]

However, other units may be used by scanners or due to previous processing of the DICOM file.

Units = “GML”

Another common value of the Units (0054,1001) attribute is grams per milliliter (“GML”). This indicates the activity concentration was already normalized. In most cases, normalization corresponds to body-weight normalization (BW), where the SUV voxel value equals the stored value multiplied by the rescale slope.

\[ \text{SUV}_{bw} = U \]

However, there may be scenarios where other normalization techniques were applied. Specifically, four SUV methods are compliant with the current DICOM standard and result in the unit GML. The method used can be extracted from the attribute SUV Type (0054,1006) and should be considered as BW if empty. The following factors are used, as reported by Sugawara:

• lean body mass by Morgan (SUV Type = “LBM”)

  for male patients: \(\text{LBM} = 1.10 W - 120 ({\scriptstyle \frac{W}{H}})^2\)

  for female patients: \(\text{LBM} = 1.07 W - 148 ({\scriptstyle \frac{W}{H}})^2\)

• lean body mass by James et al. (James, William Philip Trehearne, and J. C. Waterlow. Research on obesity. 1976) / Morgan (SUV Type = “LBMJAMES128”)

  for male patients: \(\text{LBM} = 1.10 W - 128 ({\scriptstyle \frac{W}{H}})^2\)

  for female patients: \(\text{LBM} = 1.07 W - 148 ({\scriptstyle \frac{W}{H}})^2\)

• lean body mass by Janmahasatian (SUV Type = “LBMJANMA”)

  \(\text{BMI} = \scriptstyle \frac{W}{(H \, 10^{-2})^2}\)

  for male patients: \(\text{LBM} = \scriptstyle \frac{ 9270 W}{6680 + 216 \text{BMI}}\)

  for female patients: \(\text{LBM} = \scriptstyle \frac{9270 W}{8780 + 244 \text{BMI}}\)

• ideal body weight (SUV Type = “IBW”)

  for male patients: \(\text{IBW} = 48.0 + 1.06 (H - 152)\)

  for female patients: \(\text{IBW} = 45.5 + 0.91 (H - 152)\)

For all formulas, W is weight in kg, and H is height in cm. Additionally, the following attributes are required:

• Patient’s Size (0010,1020) - Height of the patient in m

• Patient’s Sex (0010,0040) - Sex of the patient - M (male), F (female) or O (other)

In turn, this factor (LBM or IBW) can be used for normalization:

\[ \text{SUV}_{\mathrm{factor}} = \frac{\scriptstyle A_c \, \text{factor} \, 10^{3}}{\scriptstyle D} \]

To get the SUV_bw values, the stored voxel values can be divided by the corresponding factor and multiplied by the patient’s weight and the rescale slope of the slice:

\[ \text{SUV}_{bw} = \frac{\scriptstyle U \, W}{\scriptstyle \text{factor} \, 10^{3}} \]

There is currently no consensus on the computation of LBM and IBW when Patient’s Sex is specified as “O”. Using the mean value of the sex-specific factors appears to be a reasonable and simple solution.

Units = “CM2ML”

The unit square centimeters per milliliter (Units=“CM2ML”) corresponds to images normalized to body surface area (BSA) using the Du Bois formula:

\[ \text{BSA} = 0.007184 \, H^{0.725} \, W^{0.425} \]

where H is height in cm, and W is weight in kg. The SUV Type (0054,1006) should be set to “BSA” in this case. SUV_bsa is calculated as:

\[ \text{SUV}_{\mathrm{bsa}} = \frac{\scriptstyle A_c \, \text{BSA} \, 10^{4}}{\scriptstyle D} \]

SUV_bw can be backcomputed analogically to the previous section:

\[ \text{SUV}_{bw} = \frac{\scriptstyle U \, W}{\scriptstyle \text{BSA} \, 10^{4}} \]

Units = “CNTS”

Philips scanners often store the units counts (Units=“CNTS”). In this case, scale factors may be provided for converting the stored values to activity concentration or directly to SUV values. SUV scale factor allows direct conversion of the voxel values to SUV_bw. The formula is mentioned in Philips conformance statements (here, for Philips Ingenuity):

(7053,1000) DS SUV Scale Factor: This value only applies when Units (0054,1001) is equal to CNTS. The SUV Scale Factor is used to convert the voxel data from counts to an SUV value. This is done by using the formula SUV Value = ((SV x m) + b) x f, where: SV = original stored voxel value, m = Rescale Slope (0028,1053), b = Rescale Intercept (0028,1052), f = SUV Scale Factor (7053, 1000). If the SUV Scale Factor is 0.0, then the voxel data cannot be converted from counts to an SUV value.

The Activity Concentration Scale Factor allows the conversion of the voxel values to Bq/ml, as explained in the conformance statement (here, for Philips Ingenuity):

(7053,1009) DS Activity Concentration Scale Factor: This value only applies when Units (0054,1001) is equal to CNTS. The Activity Concentration Scale Factor is used to convert the voxel data from counts to Activity Concentration (in Bq/ml). This is done by using the formula Activity Concentration Value = ((SV x m) + b) x f, where: SV = original stored voxel value, m = Rescale Slope (0028,1053), b = Rescale Intercept, (0028,1052), f = Activity Concentration Scale Factor (7053, 1009). If the Activity Concentration Scale Factor is 0.0, then the voxel data cannot be converted from counts to Activity Concentration.

In case none of the factors is provided, counts (CNTS) can be converted to counts per second (CPS) by dividing by the actual frame duration in seconds, i.e., the Actual Frame Duration attribute (0018,1242) value divided by 1000.

Units = “CPS”

Counts per second (CPS) can be converted to activity concentration (Bq/ml) based on whether the image is dose calibrated or not, which can be derived from the presence of “DCAL” in the attribute Corrected Image (0028,0051):

1. in case the image is not dose calibrated, by multiplying by the Dose Calibration Factor (0054,1322) and dividing by voxel volume in ml - however, in most cases, the factor may be unknown;

2. in case the image is already dose calibrated, by dividing by voxel volume in ml:

\[ \text{SUV}_{bw} = \frac{\scriptstyle U \, W}{\scriptstyle x^{2} \, z \, 10^{-3}} \]

where x is the Pixel Spacing (0028,0030) and z is the Slice Thickness (0018,0050), which show the dimensions of a single voxel in mm.

Importantly, it must be verified that all required image corrections have been applied, as this unit is often associated with uncorrected image data.

Other units

For other units, the conversion to activity concentration or SUV_bw is impossible or unclear.

That includes, above all, following units:

• PROPCNTS - proportional to counts, not mentioned in conformance statements, conversion is not clear

• PROPCPS - proportional to counts per second, mentioned by Siemens for “non-quantitative, non-attenuation corrected images”, conversion is not clear

• 1CM - 1/centimeter, used for mu maps

Metadata analysis

The units used were highly dependent on scanner manufacturer (see Table 2):

Table 2: Frequency of unique Units (0054,1001) value by manufacturer

	BQML	CNTS	CPS	GML	PROPCNTS
Total	2768	152	120	33	33
GE	1670	0	120	23	33
Siemens/CTI/CPS	952	0	0	0	0
Philips	111	152	0	10	0
Unknown	35	0	0	0	0

There were no cases with a non-NA value in the SUV Type attribute.

Regarding counts as units and their conversion, 146 series (93.4%) with counts units had at least one scale factor provided - 136 series (89.5%) contained both factors and 6 series (3.9%) only the SUV scale factor. The Activity Concentration Scale Factor never appeared without the SUV scale factor. 10 series (6.6%) stored none of the factors. None of these 10 series was dose calibrated and the Dose Calibration Factor was never present which would disable SUV conversion.

The Philips conformance statements don’t specify whether the type of SUV acquired by applying SUV scale factor and none of the analyzed series had the SUV Type (0054,1006) stored. However, in all 136 series with both factors present, the SUV scale factor corresponded numerically to SUV_bw, as using the Activity Concentration Scale Factor and Patient’s Weight lead to practically the same SUV as using the SUV scale factor (mean relative paired difference of 0.05%).

There were 120 series with Units = “CPS”, all of them were dose calibrated and decay corrected.

Digital reference object

We synthesized multiple DROs verifying a standardized conversion to SUV_bw:

• DRO00 with Units = BQML and Decay Correction = START (baseline DRO)

  Possible issues:

  • unit BQML is not implemented.

• DRO20 Units = GML (corresponding to SUV_bw)

  Possible issues:

  • unit GML is not implemented;
  • SUV Type is required (while should be considered “BW” by default)
  • Rescale Slope is not applied;
  • other corrections are applied.

• DRO21x Units = GML (corresponding to SUV LBMJAMES128)

  • three variations: DRO211 (Patient’s Sex = “M”), DRO212 (Patient’s Sex = “F”), DRO213 (Patient’s Sex = “O”)

  Possible issues (not considering previously mentioned):

  • SUVlbm is not implemented;
  • incorrect formula for LBM is used (instead of LBMJAMES128 DICOM standard);
  • LBM unit is not changed to g;
  • patient’s weight, height or sex are not extracted correctly;
  • SUV type is not extracted correctly
  • strategy for Patient’s Sex = “O” is not implemented.

• DRO22x Patient’s Sex = “O” (other) with Units = GML (corresponding to SUV IBW)

  • three variations: DRO221 (Patient’s Sex = “M”), DRO222 (Patient’s Sex = “F”), DRO223 (Patient’s Sex = “O”)

  Possible issues (not considering previously mentioned):

  • SUVibw is not implemented.

• DRO23 Units = CM2ML (corresponding to SUVbsa)

  Possible issues (not considering previously mentioned):

  • SUVbsa is not implemented;
  • incorrect formula for BSA is used;
  • BSA unit is not changed to cm2.

• DRO24 Units = CNTS using Philips SUV scale factor

  Possible issues (not considering previously mentioned):

  • unit CNTS is not implemented;
    
  • SUV scale factor is not extracted or applied correctly.

• DRO25 Units = CNTS using Philips activity scale factor

  Possible issues (not considering previously mentioned):

  • activity scale factor is not extracted or applied correctly.

• DRO26x PatientSex = “O” (other) with Units = GML (corresponding to SUV LBMJANMA)

  • three variations: DRO261 (PatientSex = “M”), DRO262 (PatientSex = “F”), DRO263 (PatientSex = “O”)

  Possible issues (not considering previously mentioned):

  • SUVlbmjanma is not implemented.

Note: The strategy with SUV Type = “LBM” is considered obsolete and currently not covered by the digital reference objects.

Recommendations

The Units (0054,1001) attribute must be present and have one of these values: “BQML”, “GML”, “CM2ML”, “CNTS”, “CPS”; otherwise, SUV computation is not possible.

Justification: The conversion formula depends on Units.

When the Units (0054,1001) are set to “GML” and the SUV Type (0054,1006) is set to “BW” or is empty or absent, the stored voxel values can be directly converted to SUV_bw by multiplying by the Rescale Slope (0028,1053).

Justification: In this case, the voxel values correspond to units SUV_bw.

When the Units (0054,1001) are set to “GML” and the SUV Type (0054,1006) is present and has one of the following values: “LBM”, “LBMJAMES128”, “LBMJANMA”, “IBW” and the Patient’s Weight (0010,1030) and the Patient’s Size (0010,1020) have a positive, non-zero value and the Patient’s Sex (0010,1040) has one of the values “M”, “F”, or “O”, the stored voxel values can be directly converted to SUV_bw using the above mentioned formulas. When the Units (0054,1001) are set to “CM2ML” and the SUV Type (0054,1006) is set to “BSA”, and the Patient’s Weight (0010,1030) and the Patient’s Size (0010,1020) have a positive, non-zero value, the stored voxel values can be converted directly to SUV_bw using the above-mentioned formulas.

Justification: In this case, the voxel values correspond to another SUV type and have to be converted to SUV_bw.

When the Units (0054,1001) are set to “CNTS” and the Activity Concentration Scale Factor (7053,1009) is present and has a positive, non-zero value and the Manufacturer (0008,0070) attribute contains “PHILIPS”, the stored voxel values can be converted to Bq/ml by multiplying by the Rescale Slope (0028,1053) and this factor. If it is absent, negative, or zero and the SUV Scale Factor (7053,1000) is present and has a positive, non-zero value and the Manufacturer (0008,0070) attribute contains “PHILIPS” and the SUV Type (0054,1006) is “BW” or empty or absent, the stored voxel values can be directly converted to SUV_bw by multiplying by the Rescale Slope (0028,1053) and this factor.

Justification: These factors are provided by Philips to allow conversion to Bq/ml and SUV_bw.

When the Units (0054,1001) are set to “CNTS” and the Activity Concentration Scale Factor (7053,1009) is absent or empty or negative or zero and the SUV Scale Factor (7053,1000) is absent or empty or negative or zero or the Manufacturer (0008,0070) attribute does not contain “PHILIPS” and the attribute Corrected Image (0028,0051) contains “DCAL” and the voxel volume can be computed and is has a positive, non-zero value, the stored voxel values can be converted to Bq/ml by multiplying by the Rescale Slope (0028,1053) and dividing by the voxel volume in ml and Actual Frame Duration (0018,1242) in seconds. When Units (0054,1001) are set to “CPS” and the attribute Corrected Image (0028,0051) contains “DCAL”, and the voxel volume can be computed and has a positive, non-zero value, the stored voxel values can be converted to Bq/ml by multiplying by the Rescale Slope (0028,1053) and dividing by the voxel volume in ml.

Justification: Counts / s can be converted to Bq / ml by multiplying by the dose calibration factor and dividing by voxel volume.

If the Units (0054,1001) are not “BQML” and the conversion to SUV_bw or Bq/ml is not possible, SUV computation is not possible. If the Units (0054,1001) are “BQML” or the values are converted to Bq/ml, conversion from Bq/ml to SUV_bw is required.

Patient’s weight (W)

Background

Patient’s Weight (0010,1030) is often manually entered, which can result in missing values or incorrect units. The correct DICOM unit is kilograms, values equal to or greater than 1000 indicate that the weight was entered in grams. For the SUV formula, the value must be converted to grams. Furthermore, patient weight may occasionally be incorrectly stored in a different DICOM attribute, particularly Patient’s Size (0010,1020).

Metadata analysis

In our dataset, no Patient’s Weight values were equal to or greater than 1000. In 55 (1.8%) series, there was no value.

Digital reference object

–

Recommendations

If conversion from Bq/ml to SUV_bw is required, the attribute Patient’s Weight (0010,1030) should be checked that it is present and has a positive, non-zero value. Otherwise, SUV computation is not possible.

Justification: Patient’s Weight is required for conversion to SUV.

The Patient’s Weight (0010,1030) values equal to or greater than 1000 should be interpreted as grams while values lower than 1000 should be interpreted as kilograms.

Justification: Patient’s Weight may occasionally be stored in grams. A threshold of 1000 was selected pragmatically, as values equal to or greater than 1000 kg or below 1000 g are highly unlikely in clinical or phantom PET imaging. Note that the SUV formula requires weight in grams while the aforementioned LBM/IBW/BSA formulas expect weight in kilograms.

Radiopharmaceutical dose (D)

Background

The Radionuclide Total Dose (0018,1074) DICOM field records the total administered dose of the radionuclide (within the administered radiotracer). It is another attribute that is typically recorded manually, which may result in missing values or incorrect units. Furthermore, the unit may not be standardized as different Information Object Definitions (IODs) allow either Bq or MBq. The equation expects the unit to be Bq.

For SUV conversion, the administered dose must correspond to the time at which the voxel values occurred. This requires correcting the administered dose for radioactive decay of the radionuclide between the time of administration and the corresponding reference time point. Therefore, it is essential to know whether the images (voxel-values) were decay-corrected and, if so, to which time point. This information is provided by the Decay Correction (0054,1102) attribute, which can take one of three values (“ADMIN,” “START,” or “NONE”). As shown below, the subsequent correction steps depend on this attribute.

SUV is independent of the time point to which the images were corrected. Therefore, the Decay Correction (0054,1102) is not relevant when Units are “GML” or “CM2ML” or when the image can be directly scaled to SUV without considering the administered dose (Units are “CNTS” and SUV Scale Factor (7053,1000) is applied).

Decay correction = ADMIN

The value “ADMIN” indicates that the PET image data were decay-corrected to the time of radiotracer administration. In this case, no additional correction of the administered dose is required, because it already represents the dose at the time point to which the images were decay-corrected. The formula is scanner-independent and can be written as:

\[ \mathrm{SUV}_{bw} = \frac{\scriptstyle U \, W}{\scriptstyle D_\mathrm{adm}} \]

Decay correction = START

In most cases, the decay correction attribute is set to “START”, implying the series was decay corrected to a reference time representing the start of the PET image acquisition. The administered dose has to be adjusted to account for decay between administration and the reference time, using the formula: \(D = {D}_{adm} e^{-\lambda (t_\mathrm{ref}-t_\mathrm{adm})}\). Then, SUV can be computed using the formula:

\[ \mathrm{SUV}_{bw} = \frac{\scriptstyle U \, W}{\scriptstyle D_\mathrm{adm} \, e^{\left(-\lambda (t_\mathrm{ref}-t_\mathrm{adm}) \right)}} \]

where \(D_{adm}\) the administered dose of the radionuclide (0018,1074) in Bq, \(t_{ref}\) the reference time, \(t_{adm}\) the radiopharmaceutical administration time (0018,1072). \(\lambda\) represents the decay constant for the radionuclide and is computed as \(\lambda = \frac{\ln(2)}{T_{1/2}}\) where \(T_{1/2}\) is the radionuclide half life (0018,1075).

Determining the reference time:

Unfortunately, there is no single attribute reliably storing the reference time across scanners of different vendors. Moreover, the method for determining reference time is not obvious from conformance statements. A significant effort to determine a universal procedure was marked by The Quantitative Imaging Biomarkers Alliance (QIBA). In their vendor-neutral pseudo-code, they recommend to primarily use Series Date (0008, 0021) and Series Time (0008, 0031). Since they acknowledge this attribute may be often modified during processing and shifted forward, they discourage from using it when it is later than Acquisition Date (0008,0022) and Acquisition Time (0008,0032). However, as shown by Fritsak et al. and observed in our data, the Series Time may be shifted by a smaller margin or in the opposite direction, which may be difficult to recognize. In these cases, the voxel values no longer correspond to the Series Time and this leads to notable SUV computation errors.

On the other hand, Siemens and GE store the reference times in their private tags, which often remains present in the DICOM metadata, and there seem to be no signs of the time component being modified during post-processing. GE state in conformance statements of some of their models (e.g., Discovery ST/ RX/ STE) that, unlike some other models (e.g., Discovery 710/610 or Optima 560), the value of the private attribute PET scan_datetime (0009,100D) is used for setting the Series Date and Time. Furthermore, in the 2009 response to QIBA by GE, this private attribute is mentioned to be used for dose correction in case of processed images, where SeriesTime > AcquisitionTime. QIBA reflected this by recommending the use of this private datetime as the first alternative to Series Date and Time. Analogously, Siemens correct the images to the time stored in their private tag Decay Correction DateTime (0071,1022), as can be seen in the conformance statements: This attribute is not used in the QIBA strategy.

Even though the private tags appear to be the most reliable source of the reference time, there is no such attribute for Philips, and the private tags are often missing. For the remaining cases, where Series Date and Time cannot be trusted and the private tags are missing, QIBA recommends using the earliest Acquisition Time (0008,0032). The Acquisition Time, unlike the Series Time, represents the start of image acquisition for the actual frame / slice. The earliest Acquisition Time within each series should be identical to the Series Time. However, there are multiple reasons why this strategy may have limited generalizability: One such case, as already acknowledged by QIBA, are multi-injection protocols; other common pitfalls are cropped scans or scans where the first bed position was skipped. Such scenarios require slice-wise solutions. Therefore, we believe the Acquisition Time can be used only in slices where it is equal to the Series Time.

The last option mentioned by QIBA is a reference time backcomputation. A single formula - using acquisition time, frame reference time, and actual frame duration - was provided by the alliance. However, Fritsak et al. have described the differences in the reference times used by GE, Siemens and Philips. Siemens and Philips correct the images to a reference time one frame reference time before the time the values occurred - i.e., follow the equation described in the pseudocode: \[ \text{t}_{ref} = \text{t}_{acq} + \text{T}_{ave} - {\Delta{t}} \] where \({t}_{acq}\) is the acquisition time, extracted from Acquisition Time (0008,0032), \(\Delta{t}\) is the frame reference time, computed as Frame Reference Time (0054,1300) converted to seconds, and \({T}_{ave}\) is the average count rate time in seconds. The \({T}_{ave}\) is implementation- and radionuclide-dependent. In the majority of clinically-relevant cases, \({T}_{ave}\) is be computed as: \[ \text{T}_{ave} = \frac{1}{\lambda} \text{ln} \frac{(\lambda \text{T})}{1 - e^{-\lambda \text{T}}} \] where \(\lambda = \frac{\ln(2)}{T_{1/2}}\), and \(T\) is the actual frame duration, computed as Actual Frame Duration (0018,1242) converted to seconds.

In contrast, GE scanners correct the images to another time point, which is one frame reference time before the stored Acquisition Time:

\[ \text{t}_{ref} = \text{t}_{acq} - {\Delta{t}} \]

As we lack PET scans from other manufacturers, no formula can be provided for these manufacturers.

Decay correction = NONE

Another scenario involves PET images that have not been decay-corrected, with Decay Correction set to “NONE”. In this case, we must ensure that each image (voxel values) as well as the radionuclide dose is corrected to the same reference time point. The most straightforward would be correcting the dose to the measurement time (the time the values occurred) separately for each slice:

\[ \mathrm{SUV}_{bw} = \frac{\scriptstyle m \, P \, W}{\scriptstyle D_{\mathrm{adm}} \mathrm{e}^{- \lambda (t_{\mathrm{acq}} + {T}_{ave} - t_{\mathrm{adm}})}} \]

where the SUV_bw can be computed for each image directly from the values in the DICOM attributes. This method may lead to minor inaccuracies due to different implementations of the average count rate time \({T}_{ave}\) between scanner manufacturers. Once again, it must be checked that other corrections have been applied to the image.

Metadata analysis

There were 208 scans (6.7%) with a missing value of the Radionuclide Total Dose (0018,1074) attribute. Two scans (0.1%) had a negative value of this attribute. In 2870 scans with fluorine-18 as radionuclide, the Radionuclide Total Dose values ranged between 203.5 and 3.18210^09. Two distinct peaks at approximately 4 × 10² and 4 × 10⁸ suggest that the administered dose was stored in MBq or Bq, respectively (see Fig. 2). There were 45 series (1.6%) with Radionuclide Total Dose lower than 10^4. All were from the ACRIN-NSCLC-FDG-PET collection. There were 28 scans with other radionuclides, the Radionuclide Total Dose ranged between 2 10^7 and 1.4541 * 10^9.

Fig. 2: Histogram of Radionuclide Total Dose (DICOM tag 0018,1074) values for FDG-PET series.

Regarding the dose correction, most series (N=3337, 97.1%) were decay-corrected to the scan start time (Decay Correction = “START”), while there were no series decay-corrected to the administration time (Decay Correction = “ADMIN”). The remaining series were not decay corrected (Decay Correction = “NONE”, N=97, 2.8%). More than half of series (N=2040, 59.4%) contained more than one Acquisition Time (0008,0032) values.

Out of 3337 series with Decay Correction = “START”, in 189 cases (5.7%), Series Time was later than Acquisition Times in at least one slice. In 168 series (5.0%), the Series Time was before Radiopharmaceutical Administration Time.

The GE private scan datetime was present in 508 series (25.2% of GE series) and 111,797 slices. The Acquisition Time was identical to the Series Time in 44,022 slices (39.4%) and in all of these slices (100.0%), this time was also equal to the time component of the private scan datetime. Out of the remaining 67,775 slices, the GE formula led to time equal to the private scan time in 62,945 cases (92.9%). The remaining 4,830 slices (4.3%) from 84 series, where neither the Acquisition Time was equal to Series Time and the private scan time nor the GE formula led to the private scan time, represent the fraction of slices where the proposed strategy would fail if the private scan datetime was removed. The median offset from the private scan time was 422 s (IQR: 243-423 s). However, the majority (69.7%) belonged to the HN-HGJ cohort of the HEAD-NECK-PET-CT collection, where all scans have the private scan datetime provided. In contrast, the Series Time differed from the private scan time in 16,321 slices (14.6%) with both attributes present with a median offset of 7150 s (IQR: 2581-10747 s).

The Siemens private scan datetime was present in 168 series (25.0% of Siemens series) and 54,090 slices. The time was always (100.0%) identical to the Series Time. Acquisition Time was equal to the Series Time in 3,054 of these slices (5.6%). In all these slices (100.0%), this time was also equal to the time component of the private scan datetime. Out of the remaining 51,036 slices, the Siemens formula led to time equal to the private scan datetime in all cases (100.0%).

There were 2,288 series (385,445 slices) from Siemens, GE, or Philips without any private scan datetime tags. Out of these, the proposed strategy led to more than one reference time per series in only 2 series (0.1%). Furthermore, the resulting reference time was never after the Acquisition Time (0.0%). in contrast, Series Time was after Acquisition Time in 10,594 slices (2.7%) from 92 series.

There were 375 series where the Manufacturer (0008,0070) attribute did not contain any of the names of the three previously mentioned scanner manufacturers. In 279 (74.4%) of these series, the strategy led to one reference time which was before acquisition time in all series. These included all series with values “Codonics” (N=2), “CTI / MIMvista” (N=1), or “PixelMed” (N=33) in the Manufacturer (0008,0070) attribute. The Siemens/Philips formula was used in case the Acquisition Time was not equal to the Series Time. In series with Manufacturer = “CPS”, the strategy led to multiple reference times in 96/192 (50.0%) cases.

The DecayFactor (0054,1321) attribute value corresponded to correction by the Frame Reference Time for all Siemens scans (100.0%). For GE, it corresponded to the sum of Frame Reference Time and T_ave in 97.5% of slices. That shows that for both manufacturers, the measured values (values before decay correction) appeared at the time Acquisition Time + T_ave. For Philips, the Decay Factor was always set to 1.0 and cannot be used. For CPS, the Decay Factor corresponded to the sum of Frame Reference Time and T_ave in 93.6% of slices and to the Frame Reference Time in 6.4% of slices.

Digital reference object

Following DROs for decay correction alternatives were created:

• DRO30 dose in MBq

  Possible issues (not considering previously mentioned):

  • the dose is considered to be in Bq.

• DRO31 Decay Correction = “ADMIN”

  Possible issues (not considering previously mentioned):

  • Decay Correction == ADMIN is not implemented.

• DRO32x Decay Correction = “START”, private tags absent and Series Time != Acquisition Time

  • three variations for manufacturers: DRO320 (Siemens), DRO321 (GE), DRO322 (Philips)

  Possible issues (not considering previously mentioned):

  • dose is corrected to the Series Time or to the Acquisition Time;
  • the vendor-specific reference time formulas are not implemented.

• DRO33x Decay Correction = “START”, private tags present

  • two variations for manufacturers: DRO330 (Siemens), DRO331 (GE)

  Possible issues (not considering previously mentioned):

  • GE / Siemens private scan datetime are not implemented.
  • reference time formulas are used instead of the private scan datetime

• DRO34x Decay Correction = “NONE” + multiple values Acquisition Time

  • three variations for manufacturers: DRO340 (Siemens), DRO341 (GE), DRO342 (Philips)

  Possible issues (not considering previously mentioned):

  • Decay Correction == NONE is not implemented;
  • incorrect formula for voxel value decay correction or dose correction is used;
  • the voxel values and the administered dose are not corrected to the same time point.

• DRO35x Decay Correction = “START”, private tags absent and Series Time != Acquisition Time

  • three variations for manufacturers: DRO350 (Siemens), DRO351 (GE), DRO352 (Philips)

  Possible issues (not considering previously mentioned):

  • Acquisition Time is not used when Acquisition Time = Series Time

Recommendations

If conversion from Bq/ml to SUV_bw is required, the Radionuclide Total Dose (0018,1074) attribute should be present and have a positive, non-zero value; otherwise, SUV computation is not possible.

Justification: Radionuclide Total Dose is required for conversion of Bq/ml to SUV.

Radionuclide Total Dose (0018,1074) values higher than zero and lower than 10⁴ indicate storage in MBq and should be converted to Bq.

Justification: Radionuclide Total Dose may be occasionally stored in MBq and must be converted to Bq. A threshold of 10⁴ was selected based on empirical observations that doses below/ equal to 10⁴ Bq or above 10⁴ MBq are highly unlikely.

If conversion from Bq/ml to SUV_bw is required, the Decay Correction (0054,1102) attribute should be checked that it is present and has one of these values: “ADMIN”, “START”, “NONE”; otherwise, SUV cannot be computed.

Justification: Decay Correction method applied to the images must be known for conversion to SUV.

In images where conversion from Bq/ml to SUV_bw is required and Decay Correction (0054,1102) is set to “ADMIN” do not require dose correction and the conversion formula is independent of the Manufacturer (0008,0070) attribute value.

Justification: Scans with Decay Correction = “ADMIN” contain images corrected to the time point when the radionuclide was administered and its dose does not need to be corrected.

If conversion from Bq/ml to SUV_bw is required and Decay Correction (0054,1102) is set to “START” or “NONE”, dose correction is required.

Scans with Decay Correction = “START” or “NONE” store values at a different time point and the dose can be corrected to this time point. Another option is decay-correcting the images to the radionuclide administration or correcting both the dose and image to another time point. However, the selected method has no effect on SUV_bw and other methods are not implemented in these recommendations.

In images where dose correction is required and Decay Correction (0054,1102) is set to “START” and the Manufacturer (0008,0070) attribute contains “SIEMENS”, the dose should be corrected to the time stored in the private scan datetime (0071,1022) if it is present and has a non-negative value.

Justification: The private scan datetime is used by Siemens scanners for decay correction and. Based on our observations, it represents the most reliable source of the reference time. In contrast, Series Time and the earliest Acquisition Time (0008,0032) within a series show lower reliability, as demonstrated above, and are therefore not recommended.

In images where dose correction is required and Decay Correction (0054,1102) is set to “START” and the Manufacturer (0008,0070) attribute contains “GE”, the dose should be corrected to the time stored in the private scan datetime (0009,100D) if it is present and has a non-negative value.

Justification: The private scan datetime is used by GE scanners for decay correction. Based on our observations, it represents the most reliable source of the reference time.

If dose correction is required and Decay Correction (0054,1102) is set to “START” and the previous two conditions are not met and the Acquisition Time (0008,0032) in seconds is equal to the Series Time (0008, 0031) in seconds and the Manufacturer (0008,0070) attribute contains “SIEMENS” or “GE” or “Philips” and the Acquisition Time (0008,0032) has a non-negative value, the dose should be corrected to the Acquisition Time (0008,0032).

Justification: This option is used mainly for single-bed scans, first bed positions of multi-bed scans, and also accounts for the common pitfall of a constant Acquisition Time and variable Frame Reference Times within some multi-bed scans (observed in some GE and Siemens scans).

If dose correction is required and Decay Correction (0054,1102) is set to “START” and the previous three conditions are not met and the Manufacturer (0008,0070) attribute contains “SIEMENS” or “Philips” and the Acquisition Time (0008,0032) has a non-negative value and the Frame Reference Time (0054,1300) has a non-negative value and the Actual Frame Duration (0018,1242) has a positive, non-zero value, the dose should be corrected to the time one Frame Reference Time (0054,1300) in seconds before the time the pixel values in the image occurred, i.e., Acquisition Time (0008,0032) + T_ave.

Justification: This option is used mainly for multi-bed scans and is based on conformance statements (Siemens), experiments (Siemens, Philips), vendor responses (Siemens), as well as indirect evidence, such as the values of the Decay Factor (0054,1321; Siemens) and Siemens private scan datetime (0071,1022; Siemens).

If dose correction is required and Decay Correction (0054,1102) is set to “START” and the previous four conditions are not met and the Manufacturer (0008,0070) attribute contains “GE” and the Acquisition Time (0008,0032) has a non-negative value and the Frame Reference Time (0054,1300) has a non-negative value, the dose should be corrected to the time one Frame Reference Time (0054,1300) in seconds before the Acquisition Time (0008,0032).

Justification: This option is used mainly for multi-bed scans and is based on conformance statements, experiments, vendor responses, as well as indirect evidence, such as the values of the Decay Factor (0054,1321; GE) and GE private scan datetime (0009,100D; GE).

If dose correction is required and Decay Correction (0054,1102) is set to “NONE” and the Manufacturer (0008,0070) attribute contains “SIEMENS” or “GE” or “Philips”, the dose can be corrected to the time the voxel values correspond to. In that case, an identical, aforementioned equation can be used for all these cases.

Justification: For Siemens, Philips, and GE, the voxel values of non-decay-corrected images should always correspond to values at the time Acquisition Time + T_ave so the dose can be corrected to this time point, which may differ between slices / frames.

Dose correction and all other computations shall be applied on a per-slice basis, using only attributes associated with the respective slice, and shall not depend on values from other slices.

Justification: Each slice contains sufficient information for SUV computation. Scan-level approaches may introduce errors.

The manufacturer names should be recognized regardless of capitalization or the presence of additional words or characters.

Justification: Different variants of manufacturer names may appear in DICOM headers; for example, “GE MEDICAL SYSTEMS”, “GE MEDICAL SYSTEMS / MIMvista” and “GEMS” all refer to GE.

For cases where dose correction is required and the Manufacturer (0008,0070) attribute is absent or empty or not recognized as one of the manufacturers “SIEMENS”, “GE”, or “Philips”, or as a combination of these manufacturers, we cannot recommend any strategy. In such cases, SUV computation cannot be verified.

Justification: Different manufacturers may implement reference time (the time to which voxel values are decay-corrected) and measurement time (the time the original values occurred) differently. For CPS scanners, significant inconsistencies were observed. Currently, we lack sufficient data and technical information for other manufacturers to provide reliable guidance.

Radiopharmaceutical start time (t_adm)

Background

Originally, radiotracer administration time was stored in the DICOM tag Radiopharmaceutical Start Time (0018,1072). However, this attribute lacks the information about the date, therefore, was deprecated and Radiopharmaceutical Start DateTime (0018,1078) should be used for storing the information.

Clinical PET scans can be acquired on the following day after radiotracer administration, particularly when the injection occurs in the evening. Although Radiopharmaceutical Start DateTime includes the date, this information is often altered during post-processing, for example due to anonymization. Comparing Acquisition Time with Radiopharmaceutical Start Time therefore appears to be a more reliable approach for identifying scans acquired on the day following radiotracer administration. Another situation in which Acquisition Time may precede Radiopharmaceutical Start Time is dynamic imaging, where radiotracer administration occurs during the acquisition. However, such cases are typically characterized by short delays, and administration is unlikely to occur more than one hour after the start of acquisition.

Metadata analysis

Among the data examined, 1869 series (60.2%) included only the Radiopharmaceutical Start Time without an associated date, 1031 series (33.2%) included both values, and 205 series (6.6%) included neither. There was only 1 case in which only Radiopharmaceutical Start DateTime was present, which may be because the other attribute was deprecated only recently.

Out of 1032 series with both attributes available, the Radiopharmaceutical Start Time and the time component of Radiopharmaceutical Start DateTime differed in 13 series (1.3%), but the difference was always less than 60 seconds. In these series, the Radiopharmaceutical Start Time seemed to be truncated to whole minutes.

Regarding the date component of the Radiopharmaceutical Start Datetime attribute, it differed from the Acquisition Date in 212 out of 1032 series (20.5%) where both attributes were available. The most common offset was the Radiopharmaceutical Start Date one day after Acquisition Date (101 cases, 47.6%). The Radiopharmaceutical Start DateTime differed from the Series Date in 217 out of 1032 cases (21.0%). This demonstrates that these attributes are often subject to modification and therefore cannot be considered inherently reliable.

There were two series with Radiopharmaceutical Start Date one day before Acquisition Date. However, the times differed by 63 and 66 minutes, so the acquisition most likely occurred on the same day as radiotracer administration. 37 series contained slices with Acquisition Time before Radiopharmaceutical Start Time. All were dynamic scans from the “ACRIN-FLT-BREAST” collection. The highest time offset between Acquisition Time and Radiopharmaceutical Start Time was 166 seconds.

Digital reference object

Following objects were created:

• DRO40 only Radiopharmaceutical Start DateTime, no Radiopharmaceutical Start time

  Possible issues:

  • Radiopharmaceutical Start Time is required.

• DRO41 only Radiopharmaceutical Start Time, no Radiopharmaceutical Start DateTime

  Possible issues:

  • Radiopharmaceutical Start DateTime is required.

• DRO42 over midnight (started before midnight, ended after midnight, only Radiopharmaceutical Start Time provided)

  Possible issues:

  • administration time and scan time are not compared.

Recommendations

The time of radiotracer administration should be extracted from the attributes Radiopharmaceutical Start DateTime (0018,1078) or Radiopharmaceutical Start Time (0018,1072). If one of the attributes is absent, empty, or negative, the other attribute should be used. Radiopharmaceutical Start DateTime (0018,1078) should be used preferentially if both attributes are present and none of them is empty or negative. If both attributes are absent, empty, or negative and the dose correction is needed, SUV computation is not possible.

Justification: The time of radiotracer administration is required for dose and voxel value correction, i.e., as explained above, for all scans where Decay Correction is “START” or “NONE”. Both attributes store the time information. Value truncation was observed occasionally by Radiopharmaceutical Start DateTime.

There may be cases where radiotracer was administered on the day preceding the day the scan was acquired. Such cases should be recognized by a positive time difference between the time of radiotracer administration and Acquisition Time (0008,0032) higher than +3600 seconds.

Justification: In rare cases, radiotracer may be administered before image acquisition. Since the date may be modified and is not reliable, the information can be inferred from a positive difference between radiotracer administration and Acquisition Time. Since a minor positive difference may also occur in dynamic scans, a threshold of 3600 seconds is recommended to distinguish these two situations. Please note the strategy would fail in case of dynamic scans with acquisition start before midnight and radiotracer administration after midnight or if the radiotracer was administered more than 23 hours before image acquisition.

Radiopharmaceutical half-life (T_1/2)

Background

Although Fluorine-18 (F-18) is the predominant radionuclide in clinical PET imaging, other radionuclides are also used. The radionuclide information is reflected in the Radionuclide Half Life (0018,1075) attribute, which defines the physical half-life in seconds. The radionuclide information may also be encoded in the Radiopharmaceutical (0018,0031) attribute or within the Radionuclide Code Sequence (0054,0300). However, due to their frequent incompleteness and limited interpretability, these fields were omitted from the recommendations, especially considering the reliability of the Radionuclide Half Life attribute.

Metadata analysis

All series had the Radionuclide half-life provided and it always had a positive value. Most series used F-18 as radionuclide (Table 3).

Table 3: Unique values of RadionuclideHalfLife (0018,1075); half-life values were truncated.

RadionuclideHalfLife (s)	Freq	Perc	Corresponds to
598	13	0 %	N-13
1223	2	0 %	C-11
4057	20	1 %	Ga-68
6586	1201	39 %	F-18
6588	1805	58 %	F-18
23400000	65	2 %	Ge-68

Digital reference object

To check that radionuclide half-life is taken into account when SUV is computed, following DROs were constructed:

• DRO50 with Ga-68 as radionuclide, half-life provided

  Possible issues:

  • half-life is not extracted correctly from RadionuclideHalfLife;
  • radionuclide half-life is not considered within the dose decay correction.

Recommendations

In images where dose correction is required, the Radionuclide Half Life (0018,1075) attribute must be present and have a positive, non-zero value; otherwise, SUV computation is not possible.

Justification: The Radionuclide Half Life is used for dose correction.

Discussion (tbd)

In the past, efforts have been made to standardize SUV computation, most notably by the Quantitative Imaging Biomarkers Alliance (QIBA). Furthermore, digital reference objects for SUV computation have been developed; however, these only evaluate the most common scenario with Units = BQML and Decay Correction = START.

The present manual, together with the extended set of DROs, broadens the scope to cover multiple acquisition and metadata scenarios and can therefore support consistent and reproducible implementation of SUV conversion across software platforms and institutions – an aspect that is particularly critical in multi-center studies.

A limitation of this work that it solely focuses on the body-weight normalized SUV (SUV_bw). Other commonly used SUV normalizations, such as SUV_lbm, SUV_bsa, and SUV_ibw are not explicitly addressed; however, these can be computed analogously by replacing body weight with the corresponding normalization factors described above. Additionally, the analysis and resulting recommendations are partially restricted to three vendors, as data from other vendors were sparse or unavailable in both the metadata analysis and the literature.

…

Supplementary material

Suppl. Fig. 1: Flowchart showing the recommended SUV computation strategy (working vesion).

Suppl. Table 1: DICOM attributes scanned in the metadata analysis.

DICOMTag	Attribute	Vendor-specific
0008,0021	SeriesDate	No
0008,0022	AcquisitionDate	No
0008,002A	AcquisitionDateTime	No
0008,0031	SeriesTime	No
0008,0032	AcquisitionTime	No
0008,0060	Modality	No
0008,0070	Manufacturer	No
0008,103E	SeriesDescription	No
0008,1090	ManufacturerModelName	No
0009,100D	GEDecayCorrectionDateTime	GE
0009,103B	GEAdministrationDateTime	GE
0010,0010	PatientName	No
0010,0020	PatientID	No
0010,0040	PatientSex	No
0010,1020	PatientSize	No
0010,1030	PatientWeight	No
0018,1072	RadiopharmaceuticalStartTime	No
0018,1074	RadionuclideTotalDose	No
0018,1075	RadiotracerHalfLifeTime	No
0018,1078	RadiopharmaceuticalStartDateTime	No
0018,1242	ActualFrameDuration	No
0018,9701	DecayCorrectionDateTime	No
0028,1052	RescaleIntercept	No
0028,1053	RescaleSlope	No
0054,0300	RadionuclideCodeSequence	No
0054,1000	SeriesType	No
0054,1001	Units	No
0054,1006	SUVType	No
0054,1102	DecayCorrection	No
0054,1300	FrameReferenceTime	No
0054,1321	DecayFactor	No
0054,1322	DoseCalibrationFactor	No
0071,1022	SiemensDecayCorrectionDateTime	Siemens
7053,1000	PhilipsSUVScaleFactor	Philips
7053,1009	PhilipsActivityConcentrationScaleFactor	Philips

Suppl. Table 2: List of created DROs to verify SUV computation.

Section	ID	Description
default	DRO_0_0	default DRO with Units = BQML and DC = START
rescaleslope	DRO_1_0	multiple values Rescale Slope
units	DRO_2_0	Units = GML (corresponding to SUVbw)
units	DRO_2_1_0	Units = GML with PatientSex = “M” (male) corresponding to SUVlbmjames128
units	DRO_2_1_1	Units = GML with PatientSex = “F” (female) corresponding to SUVlbmjames128
units	DRO_2_1_2	Units = GML with PatientSex = “O” (other) corresponding to SUVlbmjames128
units	DRO_2_2_0	Units = GML with PatientSex = “M” (male) corresponding to SUV IBW
units	DRO_2_2_1	Units = GML with PatientSex = “F” (female) corresponding to SUV IBW
units	DRO_2_2_2	Units = GML with PatientSex = “O” (other) corresponding to SUV IBW
units	DRO_2_3	Units = CM2ML corresponding to SUVbsa
units	DRO_2_4	Units = CNTS using Philips SUV scale factor
units	DRO_2_5	Units = CNTS using Philips activity scale factor
units	DRO_2_6_0	Units = GML with PatientSex = “M” (male) corresponding to SUVlbmjanma
units	DRO_2_6_1	Units = GML with PatientSex = “F” (female) corresponding to SUVlbmjanma
units	DRO_2_6_2	Units = GML with PatientSex = “O” (other) corresponding to SUVlbmjanma
dose	DRO_3_0	dose in MBq
dose	DRO_3_1	DC = ADMIN
dose	DRO_3_2_0	DC = START but SeriesTime after AcquisitionTime SIEMENS
dose	DRO_3_2_1	DC = START but SeriesTime after AcquisitionTime GE
dose	DRO_3_2_2	DC = START but SeriesTime after AcquisitionTime PHILIPS
dose	DRO_3_3_0	DC = START but SeriesTime after AcquisitionTime private tag present SIEMENS
dose	DRO_3_3_1	DC = START but SeriesTime after AcquisitionTime private tag present GE
dose	DRO_3_4_0	DC = NONE + multiple values ACQ TIME SIEMENS
dose	DRO_3_4_1	DC = NONE + multiple values ACQ TIME GE
dose	DRO_3_4_2	DC = NONE + multiple values ACQ TIME PHILIPS and Units = CNTS
dose	DRO_3_5_0	DC = START but SeriesTime = AcquisitionTime SIEMENS
dose	DRO_3_5_1	DC = START but SeriesTime = AcquisitionTime GE
dose	DRO_3_5_2	DC = START but SeriesTime = AcquisitionTime PHILIPS
rptime	DRO_4_0	only radiopharmaceutical datetime, no RP Date and RP time
rptime	DRO_4_1	only radiopharmaceutical time, no RP datetime
rptime	DRO_4_2	Sparing midnight
halflife	DRO_5_0	Radionuclide Ga68